https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122172
--- Comment #15 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
(In reply to Hans-Peter Nilsson from comment #14)
> (In reply to Jonathan Wakely from comment #13)
> > OK, thanks, I'll get the atomicy_dir stuff cleaned up.
> >
> > I'm also testing a simpler patch to fix all the warning regressions.
>
> Not by just silencing it, I hope.
>
> I'll move forward with testing the 2+3 patch (and whatever new
> unsigned-atomic things that may have been introduced and missed) against
> current sources (r16-4259-g9242a89d088f44).
(...looking at the results, surprised by the regressions appearing to remain)
Oh, you mean:
/gccobj/cris-elf/libstdc++-v3/include/bits/shared_ptr_base.h:270: warning:
ignoring attributes on template argument '_Atomic_word' {aka 'int'}
[-Wignored-attributes]
which I guess to come from thee __int_traits instantiation (sp?) in:
constexpr _Atomic_word __max
= sizeof(long) > sizeof(_Atomic_word)
? -1 : __gnu_cxx::__int_traits<_Atomic_word>::__max;
Yeah, *that* one can be safely hacked off, somehow, presumably some cast.
I don't really know what to put there, so I guess I'll just have to stand by to
test whatever patch you think of, for cris-elf (within 24h of a patch posted to
the PR, if you need that).
